Hotelling’s T 2 control chart using a Stein-type shrinkage estimator for monitoring high-dimensional data
Summary
No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.
Summarise with VIRAT Free to try with your Science Catalogs account.Abstract
The Hotelling’s T2 chart, which tracks the process mean vector, is a commonly used tool for tracking and managing multivariate processes. Nonetheless, the T2 control chart is (i) ineffectual for monitoring processes with a restricted number of preliminary subgroups m relative to a greater number of variables p, and (ii) incalculable when p≥m due to the singularity of the sample covariance matrix. To address these computational challenges, this work proposes a modification of the classical Hotelling’s T2 control statistic using a new Stein-type shrinkage estimator of the covariance matrix. The proposed approach shrinks the eigenvalues of the sample covariance matrix, producing a well-conditioned and invertible estimate suitable for high-dimensional settings. This enables effective monitoring of the mean vector in multivariate processes using individual observations for p≥m settings. The stability of the estimator is enhanced through an adjustment parameter based on perturbation theory. The methodology incorporates two different target structures and uses an iterative procedure to optimize shrinkage intensity. We evaluate the Phase II performance of the proposed control chart using out-of-control average run length measurements. Simulation results demonstrate that this modified T2 control chart effectively detects process shifts in the p≥m paradigm.
Fields of Study & Subjects
Advanced Statistical Process Monitoring
Fault Detection and Control Systems
Statistical Methods and Applications
Publication details
- Journal / source
- Quality Technology & Quantitative Management
- Pages
- 1–30
- Publication date
- 31 March 2026
- Language
- EN